Hvplot data with inf value gives empty plot

hvPlot
1

The y-range seems to be reset to (0,1) when there is an inf value in the data. The data is actually in the plot with the inf value is removed, but you only see an empty plotting window. pandas.plot seems to handle this better. Can anyone comfirm this bug?

df = pd.DataFrame(data=np.cumsum(np.random.normal(size=(100,1)))+100,columns=['data'])
df_inf = df.copy()
df_inf.iloc[2,:] = np.inf

%%capture
ax = df_inf.plot(y='data',title='matplotlib get y-range correct')

pn.Column((df_inf.hvplot(y='data',label='contains inf value') + 
           df_inf.hvplot(y='data',label='contains inf value, manual ylims (90,110)').opts(ylim=(90,110)) + 
           df.hvplot(y=[('data','data inf removed')],label='removed inf value')).opts(hv.opts.Curve(axiswise=True,width=400,height=200)).cols(1),
           ax.figure
         )

image
image421×970 46 KB

2

I just checked your example: hv.Curve(df) does indeed return an empty plot,
though, what axis limits would you expect the function to choose for y values
when the extent is infinite?

3

@ea42gh, I had the same reasoning at first, but as you can see: it filters out the inf (third) value, just like matplotlib but then fails show reasonable axis limits for the remaining data. I’m guessing there is a solution for this.

4

Create new issue on github page: https://github.com/holoviz/hvplot/issues/553